1. Field of the Invention
The invention relates to the protection of integrated circuits from electrostatic discharge (ESD), and more particularly to a low voltage protection of CMOS circuits by a parasitic silicon controlled rectifier (SCR) which is triggered by a displacement current.
2. Description of the Related Art
The protection of integrated circuits from electrostatic discharge (ESD) is a subject which has received a lot of attention from circuit designers because of the serious damage that ESD can wreak as oxide thicknesses are reduced together with device dimensions. The seriousness of the problem is reflected in the number of articles published and U.S. patents issued. Workers in the field and inventors have proposed many solutions, many trying to solve the problem of protecting sub-micron devices while still allowing them to function unencumbered and without undue, or zero, increase of silicon real estate. The main thrust of ESD protection for CMOS devices is focused on the use of parasitic npn and pnp bipolar transistor, which together form an SCR. Unwanted as this SCR is normally, it can safely discharge dangerous ESD voltages as long as its trigger voltage is low enough to protect those CMOS devices of which it is a part. Past solutions have offered low voltage lateral SCRs (LVTSCR), modified lateral SCRs (MLSCR), PMOS-trigger lateral SCRs (PTLSCR), NMOS-trigger lateral SCRs (NTLSCR), and modified PTLSCRs and NTLSCRs.
FIG. 1 is a cross-sectional view of a typical prior art CMOS structure protected from ESD pulses by an SCR and NMOS transistor. Shown is a semiconductor wafer 10 with CMOS devices, with parasitic bipolar transistors forming an SCR, and an additional NMOS device for lowering the trigger voltage of the SCR. In a p-substrate 11 an n-well 12 is formed, and a p-channel transistor with a p.sup.+ source 14 and a p.sup.+ drain (not shown) is created. An n.sup.+ contact region 13 is formed in the n-well and together with p.sup.+ source 14 connected to a voltage supply pad 19. In p-substrate 11 an n-channel transistor with an n.sup.+ drain 15, an n.sup.+ source 16, and a gate 17 is created. The n.sup.+ drain 15 straddles p-substrate 11 and n-well 12. A p.sup.+ contact region 18, formed in p-substrate 11, is connected together with n.sup.+ source 16 to a reference voltage 20.
The steps that produce the above CMOS structure also create parasitic bipolar pnp transistor 21 between p.sup.+ source 14 (emitter), n-well 12 (base), and p-substrate 11 (collector), and parasitic bipolar npn transistor 22 between n.sup.+ source 16 (emitter), p-substrate 11 (base), and n-well 12 (collector). The base of transistor 21 is connected via n-well resistor 23 to n.sup.+ contact region 13, and the base of transistor 22 is connected via p-substrate resistor 24 to p.sup.+ contact region 18. The base of one transistor is connected to the collector of the other transistor. Resistors 23 and 24 are equivalent resistors for the intrinsic resistance of the n-well and p-substrate material. FIG. 2 is the equivalent circuit of FIG. 1 showing the interconnection of transistor 21 and 22 forming an SCR. NMOS transistor Q1 is shunted across npn transistor 22 providing the trigger for the SCR. ESD voltage pulses are shunted from pad 19 via transistors 21 and 22 to reference voltage (ground) 20.
U.S. patents relating to ESD protection are:
U.S. Pat. No. 5,754,381 (Ker) provides a modified PTLSCR and NTLSCR, and bypass diodes for protection of the supply voltage and output pad of an output buffer. The trigger voltage is the low snap-back trigger voltage of a short-channel PMOS (NMOS) device. PA1 U.S. Pat. No. 5,754,380 (Ker et al.) is similar to U.S. Pat. No. 5,754,381 above but without bypass diodes. The invention requires a smaller layout area than conventional CMOS output buffers with ESD protection. PA1 U.S. Pat. No. 5,744,842 (Ker) teaches the use of an ESD protection circuit between V.sub.dd and V.sub.ss, comprising a bipolar transistor and an n-type field-oxide device. PA1 U.S. Pat. No. 5,742,085 (Yu) discloses a low-voltage trigger protection, consisting of two SCRs and an NMOS transistor for triggering disposed between an IC bonding pad and V.sub.ss. PA1 U.S. Pat. No. 5,719,733 (Wei et al.) presents an ESD protection device using one SCR for deep submicron CMOS devices, where the structure comprises a p-well and an n-well (or p-substrate) that are separated. A ground electrode is connected to a p.sup.+ and an n.sup.+ contact region and through a polysilicon region to a gate oxide region in the n-well. The triggering voltage for snapback of the SCR is tunable between 5-11 Volts. PA1 U.S. Pat. No. 5,576,557 (Ker et al.) provides ESD protection for sub-micron CMOS devices supplying discharge paths at V.sub.dd and V.sub.ss using two LVTSCRs. In addition a PMOS device is used in conjunction with one LVTSCR and an NMOS device with the other LVTSCR. Inclusion of the PMOS and NMOS devices allows lowering of the trigger voltage to 11-13 Volt. PA1 U.S. Pat. No. 5,572,394 (Ker et al.) discloses an ESD protection circuit formed by two PTLSCRs and two NTLSCRs for protection from the four modes of ESD. The use of short-channel thin-oxide PMOS (NMOS) devices reduces the turn-on voltage of the lateral SCR to below the gate-oxide breakdown voltage of CMOS devices in the input stage. PA1 U.S. Pat. No. 5,541,801 (Lee et al.) uses three LVTSCRs which are connected between V.sub.dd, the circuit to be protected, and V.sub.ss. Each of the SCRs uses a PMOS/NMOS transistor to lower the trigger voltage. The gates of the PMOS/NMOS transistors are each in turn connected via linked terminals of trigger gates to the circuit to be protected. PA1 U.S. Pat. No. 5,452,171 (Metz et al.) describes a protection circuit using an inverter trigger device, with a voltage divider on its output, to control the amount of voltage necessary to cause latchup of the parasitic SCR. An NMOS device is disposed between the output of the trigger device and the SCR. PA1 U.S. Pat. No. 5,400,202 (Metz et al.) is similar to U.S. Pat. No. 5,452,171 above in the use of an inverter trigger device, an NMOS transistor and the parasitic SCR, but does not appear to use the voltage divider.
It should be noted that none of the above-cited examples of the related art have reduced the snapback voltage below 4 Volt. The snapback voltage of a LVTSCR with a short-channel NMOS (or PMOS) of 4 Volt is still too high to protect the internal circuit of 0.18 micron processes, because the oxide thicknesses are in the order of 32 Angstrom. The related art does not address protection for oxides that thin. What is needed is an SCR with an trigger voltage in the range of 1.2 Volt. This voltage is low enough to prevent internal device damage by ESD pulses. Instead of the breakdown of an NMOS (or PMOS) device utilizing the breakdown of n-well to p-substrate (or p-well) will be proposed.